How to Find the Inverse of a Matrix: A Complete Guide

Ever tried to solve a simple algebra problem like 5x = 10? You just divide 10 by 5 to get x = 2. You’re really “undoing” the multiplication. In the world of economics and econometrics, we deal with much bigger systems. We might have dozens of equations with dozens of unknowns, all neatly organized into special arrays of numbers called matrices. We often face a problem that looks like AX = B, where A is a matrix of known coefficients, X is a vector of unknowns we want to find (like the impact of interest rates and education on income), and B is a vector of outcomes. But there’s a catch: you can’t just “divide” by a matrix. Instead, we use a powerful concept that acts like division: the matrix inverse.

Finding this inverse is like finding a universal “undo” button for our matrix A. It’s the key that unlocks the solution to our entire system of equations. But how do we find this key? It’s not as simple as flipping a number, but it’s a systematic process that, once mastered, becomes an indispensable tool in your analytical toolkit.

So, what exactly is a matrix inverse? 🔗

Let’s start with the formal definition. For a square matrix A (meaning it has the same number of rows and columns, like 2×2 or 3×3), its inverse is another matrix of the same size, which we call A⁻¹ (read as “A-inverse”).

What makes it special is that when you multiply A by its inverse A⁻¹, in any order, you get a very special matrix called the identity matrix, or I.

So, the magic formula is: A * A⁻¹ = I and A⁻¹ * A = I

The ‘one’ of the matrix world: The identity matrix (I) 🔗

The identity matrix, I, is the matrix equivalent of the number 1. Just as 5 * 1 = 5, any matrix A multiplied by the identity matrix I is just A (A * I = A). It’s a square matrix that has 1s on its main diagonal (from top-left to bottom-right) and 0s everywhere else.

For example, the 2×2 identity matrix is:

I =
| 1 0 |
| 0 1 |

And the 3×3 identity matrix is:

I =
| 1 0 0 |
| 0 1 0 |
| 0 0 1 |

So, finding A⁻¹ is all about finding the one matrix that “cancels” A and leaves us with this clean, simple identity matrix.

The most important rule: What is a non-singular matrix? 🔗

Here’s the catch, and it’s a big one: not every square matrix has an inverse. It’s just like how the number 0 doesn’t have a reciprocal (1/0 is undefined); some matrices don’t have an “inverse.”

A matrix that has an inverse is called non-singular or invertible.
A matrix that does not have an inverse is called singular.

How do we know which is which? We need to calculate a single, powerful number that tells us the matrix’s most important properties. This number is the determinant.

The determinant: The gatekeeper to the inverse 🔗

The determinant of a square matrix, written as det(A) or |A|, is a special scalar value computed from its elements. For our purposes, its most important job is to tell us if an inverse exists.

The rule is simple:

• If det(A) ≠ 0 (the determinant is not zero), the matrix is non-singular and an inverse exists.
• If det(A) = 0 (the determinant is zero), the matrix is singular, and no inverse exists. Stop right here!

In economics, a singular matrix often means you have a problem in your model. It might mean two of your variables are perfectly correlated (e.g., measuring temperature in both Fahrenheit and Celsius), a situation called multicollinearity. The model is “broken” because the math is telling you some of your information is redundant.

Calculating the determinant for a 2×2 matrix 🔗

This is the simplest case. For a 2×2 matrix:

A =
| a b |
| c d |

The determinant is: det(A) = (a*d) – (b*c)

You multiply the main diagonal (a*d) and subtract the product of the other diagonal (b*c).

The main event: Calculating the inverse using the adjoint 🔗

Once we know our determinant is non-zero, we can proceed. The most common formula-based method for finding the inverse, especially for 2×2 or 3×3 matrices, uses a component called the adjoint.

The formula is:
A⁻¹ = (1 / det(A)) * adj(A)

This looks simple, but it has three distinct parts:

1. Find the determinant, det(A).
2. Find the adjoint, adj(A).
3. Multiply the adjoint by the scalar 1/det(A).

We already know how to find the determinant. The real work is finding the adjoint. The adjoint is the transpose of the cofactor matrix. This introduces two new terms: “cofactor” and “transpose.” Let’s break them down.

Step 1: Minors and cofactors 🔗

A minor is the determinant of a smaller matrix. To find the minor of an element, you “delete” the row and column that element is in and then find the determinant of what’s left.

A cofactor is just the minor with a specific sign (+ or -) attached to it. The sign depends on the element’s position. You use a “checkerboard” pattern of signs:

| + – + |
| – + – |
| + – + |

So, the cofactor of the element in the 1st row, 1st column is (Minor). The cofactor of the element in the 1st row, 2nd column is -(Minor).

Step 2: The cofactor matrix (C) 🔗

The cofactor matrix is a new matrix, the same size as A, where each element is replaced by its corresponding cofactor. We call this matrix aws C.

Step 3: The adjoint matrix (adj(A)) 🔗

The adjoint is the transpose of the cofactor matrix. To “transpose” a matrix, you simply swap its rows and columns. The first row becomes the first column, the second row becomes the second column, and so on.

So, adj(A) = Cᵀ (where Cᵀ is the transpose of the cofactor matrix).

Step 4: The final inverse 🔗

Once you have your adjoint matrix, you just multiply every single element inside it by the fraction 1/det(A). This is called scalar multiplication. The result is your final inverse matrix, A⁻¹.

Let’s make it real: A 2×2 example, step-by-step 🔗

That all sounds very theoretical. Let’s walk through a concrete 2×2 example. It’s much simpler than the general process suggests.

Let’s find the inverse of this matrix:
A =
| 4 7 |
| 2 6 |

Step 1: Find the determinant.
det(A) = (a*d) – (b*c) = (4 * 6) – (7 * 2) = 24 – 14 = 10
Since 10 ≠ 0, an inverse exists!

Step 2 (Shortcut for 2×2): Find the Adjoint.
For a 2×2 matrix, the adjoint has a wonderful shortcut. For our matrix A = [[a, b], [c, d]]:
adj(A) =
| d -b |
| -c a |

You swap ‘a’ and ‘d’ on the main diagonal, and you change the signs of ‘b’ and ‘c’ on the other diagonal. (This shortcut is the result of the whole minor-cofactor-transpose process, just simplified).

For our matrix A:
adj(A) =
| 6 -7 |
| -2 4 |

Step 3: Multiply by 1/det(A).
A⁻¹ = (1 / det(A)) * adj(A)
A⁻¹ = (1 / 10) *
| 6 -7 |
| -2 4 |

Now, we multiply each element inside by 1/10:

A⁻¹ =
| 6/10 -7/10 |
| -2/10 4/10 |

Which simplifies to:
A⁻¹ =
| 0.6 -0.7 |
| -0.2 0.4 |

And that’s it! We found the inverse. We could even check our work by multiplying A * A⁻¹ to see if we get the identity matrix I. (Spoiler: we do).

Handy properties of the matrix inverse 🔗

As you get comfortable with inverses, you’ll want to use their properties to save time. These are the fundamental rules that always hold true, as explained in many linear algebra texts.

The ‘socks and shoes’ rule: Inverse of a product 🔗

This is the most famous (and often-tested) property. If you have two invertible matrices, A and B, the inverse of their product is:

(AB)⁻¹ = B⁻¹A⁻¹

Notice the order is reversed. It’s called the “socks and shoes” rule because to undo getting dressed, you must take off your shoes *first* (B) and *then* your socks (A). You can’t undo it in the same order you did it.

The inverse of a transpose 🔗

The inverse of a transpose is the transpose of the inverse. It doesn’t matter which operation you do first, which is very convenient.

(Aᵀ)⁻¹ = (A⁻¹)ᵀ

The inverse of a diagonal matrix 🔗

This is the easiest one. A diagonal matrix is one with non-zero values only on the main diagonal (like the identity matrix, but with different numbers). To find its inverse, you simply invert each diagonal element individually.

If D =
| 5 0 0 |
| 0 2 0 |
| 0 0 4 |

Then D⁻¹ =
| 1/5 0 0 |
| 0 1/2 0 |
| 0 0 1/4 |

This property is extremely useful in econometrics, particularly when dealing with variance-covariance matrices of uncorrelated variables.

What do you think? How might the concept of a ‘singular’ matrix (one with no inverse) show up as a practical problem in an economic model you’re trying to build? And now that you’ve seen the 2×2 shortcut, can you trace back how it’s just a simplified version of the full cofactor and transpose method?

References

1. https://math.mit.edu/~gs/linearalgebra/ila0106.pdf
2. https://byjus.com/maths/scalar-multiplication-of-matrices/
3. https://math.libretexts.org/Bookshelves/Linear_Algebra/A_First_Course_in_Linear_Algebra_(Kuttler)/02%3A_Algebras/2.06%3A_Properties_of_the_Inverse
4. https://www.statlect.com/matrix-algebra/diagonal-matrix